Data QA Engineer

At Tive, we imagine a fully-transparent world in which everything and everyone is connected. We innovate beyond what the world thought possible-so what people hold near and dear arrives on time and in full.

We believe (and live!) four core values:

• Transparency First: We make everything visible so that everyone can operate better
• Always strive to make things simpler
• We are One Team, we have each other's backs
• Relentless iteration to optimize and improve

Whether shipped by road, rail, ocean, or air, Tive reduces shipment delays, minimizes rejected loads, and decreases theft, damage, and spoilage. Our customers count on Tive to ensure that their shipments are delivered on time and in full-because every shipment matters.

Tive's shipment-tracking platform turns device sensor data into customer-facing analytics, and increasingly into AI agents that answer questions about that data. The Data QA Engineer owns whether those outputs can be trusted. In this role you will build and run data quality for our dbt Snowflake pipelines, from raw source tables to customer-facing marts. You'll also design how we evaluate and test the AI agents built on top of that data.

Our ideal candidate has:

• 4+ years in data QA, analytics engineering or data engineering with a quality focus.
• Strong SQL. You can read a complex DBT model on your own, work out its grain and join risks, and write validation queries that size an issue.
• Hands-on DBT experience: tests, contracts, schema.yml documentation, sources and freshness, and how incremental models fail.
• Experience with a cloud data warehouse, preferably Snowflake.
• Experience designing tests with judgment: knowing which checks matter for which data, and keeping alerts rare and actionable.
• A working understanding of how LLMs and agents behave, including non-determinism, hallucination and tool use, and how to measure systems whose output isn't the same every time.
• Comfort with Git, pull request reviews and CI.
• Clear writing. You can turn a data problem into a ticket an engineer can act on, and explain quality results to people who aren't technical.

Nice to have:

• Elementary, dbt_utils, dbt_expectations or Snowflake data metric functions.
• LLM evaluation tools such as LangSmith, Braintrust, MLflow, DeepEval or Snowflake Cortex Analyst evaluations.
• Experience evaluating text-to-SQL or analytics agents, or building semantic layers.
• Airflow, and building BI dashboards (Sigma or similar).
• Using AI coding assistants to speed up QA work, such as model review, test generation and failure triage.
• Experience with IoT, sensor or logistics data.

What you’ll be doing:

• Own the data quality strategy for our Snowflake + dbt + Airflow pipeline. Decide which models get which checks based on pipeline layer and customer impact.
• Write and maintain dbt tests. Cover uniqueness, not-null, accepted values, referential integrity, timestamp order, value ranges and business logic. Use dbt built-ins, dbt_utils, dbt_expectations, unit tests and custom generic tests.
• Monitor data pipeline health. Configure source freshness and anomaly tests, and tune severity and thresholds so alerts are rare and actionable. Build and maintain dashboards on test results so the team can see data health at a glance.
• Raise the bar in development to increase data quality. Review pull requests for test coverage and data risk, improve CI testing, and get QA involved before models ship, not after.

• Build evaluation datasets for AI agents. Work with analysts and domain experts to create golden question sets with verified answers and verified SQL, covering common questions, edge cases and questions the agent should decline.
• Create agent test suites: Write scoring rubrics and metrics for answer accuracy, SQL correctness, tool and data-source choice, groundedness and latency. Run the eval suite on every change to prompts, models, semantic views or underlying dbt models, and flag answers that used to be right and now aren't.
• Analyze agent failures and monitor in production. Read traces of multi-step runs to find out why an agent picked the wrong table, filter or tool, and route the fix to the right owner (prompt, semantic layer, data model or source data). Track live quality over time and report it in terms the product and leadership teams can act on.

What does Tive offer?

• A chance to join what may very well turn out to be the most important company in your career.
• The autonomy and resources to build what you know how to build.
• Work with a committed global team that have each others back.
• Office-based or hybrid options. Your choice.
• Competitive equity to ensure all of our employees have a sense of ownership in the long-term success of Tive’s growth.

We celebrate diversity, and consider it key to our success as both a team and a company. We are proud to be an equal-opportunity employer, and we are committed to creating an inclusive environment of mutual respect for all employees.